SSD Question - Vertex 4 to Samsung 840

[Arkaign]

Hey, I just wanted to confirm what I've done to make sure it makes sense.

I had an OCZ Vertex 4 256GB SSD, worked great, but I had the opportunity to grab a Samsung 840 250GB that I couldn't pass up. It's not the Pro version, so it's TLC memory. As far as I can tell, the 840 is equal or a good bit faster in read speed, and a bit slower in writes depending on circumstance.

The only things on the SSD are Windows, games, and apps. All of my download/documents/music/etc are mapped to raid WD Black 1TBs.

I guess my question is, I shouldn't really notice much of the reduced write performance, correct? That would only come much into play with installing things (seldom), and the increased read performance should boost my daily use performance.

FWIW, I will probably use this drive until 500gb models of equal or better performance hit the $200 range. (1-2 years?)

[postmortemIA]

correct, you rarely write much data (compared to reads); unless if you are pro photo/video editor. I've just looked in Windows Task Manager, and most writes in my current session were from:
Firefox
Flash
Outlook
If these are apps that do the most writes, then there's nothing to worry about.

[Arkaign]

Thanks, that makes sense.

[Puffnstuff]

I'm waiting on that 500gb 840 to come down myself. Right now it's $349 at newegg but I'm hoping to see it under $300 before too long. I want to start replacing my spinners with ssd's.

[Coup27]

Why did you swap? You haven't gained any more capacity and there will be no noticeable difference in performance between the two outside of some useless benchmark numbers.

[bigystyle84]

Why Swap?

I have 2 x 256GB Vertex 4 in Raid 0 and they are FAST and issue free for me.

[StriperMike]

I was looking at the same two drives as an upgrade. Interested as to why you went from the seemingly better drive in the OCZ to the 840.

[Arkaign]

Originally Posted by StriperMike

I was looking at the same two drives as an upgrade. Interested as to why you went from the seemingly better drive in the OCZ to the 840.

Look at the at ssd bench comparison, my tests confirm notably better read performance, particularly with nonsequential and lots of small files in addition to faster sequential. Writes are down slightly.

[groberts101]

All this stuff boils down to is brand loyalty, pricing, and personal preference, IMO.

Like I've always said in the past.. we don't really wait for reads.. we wait for writes and that's ultimately what bogs a firmware and controller down when things get hot and heavy for a multi-tasking power user.

But if you're not a power user in the first place?.. good luck telling the difference between an ultra-low latency drive with already fast reads.. and another ultra-low latency drive with only slightly faster reads.

[Arkaign]

I don't know about that, I'm loading BF3 maps a decent bit quicker with this new drive. And with all my apps already installed and seldom installing new ones, with storage/download on large mechanical drives, this is a better fit for me. I quite liked the Vertex 4, nothing wrong with it, but this is pretty much a perfect fit.

I think it would be way different if I was writing a lot to the C drive, but my workloads with writes (VMs, Photoshop, Premiere, Solidworks, etc) are all using and needing much larger storage areas. It's not that I'm not a "power user" or whatever condescending term that is, it's that I don't need heavy write speed on the SSD.

Perhaps some just aren't aware of the considerable improvements for read-heavy (waiting for large apps to load, etc) that even just the 840 offers. It looks like I will be able to get a pair of 500 840 pros soon though, so I may be able to move a greater amount of workdata to SSDs entirely.



Those are the sections which really show the improved speeds to add up nicely. The black bar is the 840.

[TemjinGold]

There's still a heavy misconception that one would never see ANY difference in normal use between any two SSDs. I wonder how many of those people are simply saying that because they heard someone else say it. I've heard plenty how even from my G2 to a modern SSD I wouldn't see any difference. Your experience is making me more eager to search out a deal and upgrade.

[Ao1]

You are right, read performance does vary from SSD to SSD. Some of the newer (and older “out of date”) drives are maximising opportunities for read ahead to improve “real world” I/O and the 840 is one of those drives. Other drives like SF based drives don’t (currently).

The M4 and Plextor M5Pro are great performers for “real world” reads. The 840 and 840Pro are also really fast, but I have not had chance to update/ include them in the charts.



Uploaded with ImageShack.us

The M4 is let down my poor write performance, but how often are write intensive tasks relevant to a client user? Most write I/O can also be delayed by lazy writes to occur at a convenient time that will not impact performance. There is also a big difference in “real world” write performance depending on the mix of write I/O. For example the Vertex 4 comes out fastest for tasks that mainly consist of large sequential xfers, but does less well when the workload is more random.



Uploaded with ImageShack.us

[BlakeBang]

I dont think that will make a difference. The performance difference isnt significant enough and it's highly unlikely you ll notice any difference.

Respect the tests but Vertex will is a fine SSD. You aren't going to use your PC under extreme conditions of testting.

[TemjinGold]

Umm... the OP just said he notices a significant difference so why do you think he won't? Have you used these drives too or are you just saying that because that seems to be trendy on the net?

[Mfusick]

I think it placebo or perhaps a fresh install vs a bloated one....


Every new install feels faster to me even when I use the same drive...

[groberts101]

The Samsung chior can sing all they want about this drive.. but until they fix that poor speed degradation issue(extra.. extra.. read all about it).. I wouldn't touch it with a 10 foot pole for an OS volume. Because that's where the V4 spanks this drive when its actually filled up with an OS/apps/scratch data and pushed harder than just loading BF3 maps.

I mean seriously.. who doesn't fill their drives past 50% and actually scratch all that HD media to their C-drive?

Not to say that some folks can't possibly find improvements in their particular workflow(the firmware is actually tuned quite nicely for a primarily read-only storage drive).. just that this is not a multitasking monster of a drive with fast enough writes and suitable on-the-fly recovery(read as "agressive GC" which would be contradictory to lifespan preservation of its precious TLC) to suit anyone who doesn't want to move all their temp files and limit usage to heavy reads with minimal writes, is all. Good luck achieving that usage model with an OS volume on this drive though.

As for the Anand Benchmark charts?.. LOL.. fill the drive to 80% and give the 840 another shot at the V4. I've already seen the results firsthand and it ain't pretty.

And personally speaking.. I liked the previous 1.4 firmware better than the newer 1.5 but obviously OCZ felt the need to reduce benchmark performance for the tradeoff of more agressive recovery. That alone should make you wonder why such a performance oriented company would do such a thing. I'm guessing that it's mainly because they now realize that LONGTERM performance.. when people overfill their drives and still feel the need to benchmark them.. is far more important than those initial peak benchmark numbers on fresh drives. At least with the V4.. you load er' up and go without worrying about moving a damned thing off the drive until it gets past 90% or so. Which inevitably means that having to shortstroke an 840 squashes some of those savings you just found from buying TLC.

[Coup27]

Originally Posted by groberts101

The Samsung chior can sing all they want about this drive.. but until they fix that piss-poor speed degradation issue(extra.. extra.. read all about it).. I wouldn't touch it with a 10 foot pole for an OS volume.

I don't think anybody is really singing many praises about the 840, including me. I have defended the endurance of TLC NAND as it is still more than adequate for the average user, but with the 840's low sequential write speed and the market being as competitive as it is, I would probably look elsewhere from an 840 at the moment, but it depends on the build and what's on offer at the time.

Out of interest, I have not heard about the speed degradation issue. Can you point me to some information?

[Arkaign]

Originally Posted by Mfusick

I think it placebo or perhaps a fresh install vs a bloated one....


Every new install feels faster to me even when I use the same drive...

I know what you mean. In this case I used Acronis to copy it over though. Had to reactivate a couple pieces of software but it didn't take long.

[bigystyle84]

I went from an Intel 320 to a Vertex 4 and never noticed any difference for daily use. (actually using the apps, browsing, gaming, etc)

The Only thing I've really noticed is Vertex 4 is much quicker at Installing Apps, etc. Which makes sense.

I just gave my dad a Agility 4 for Christmas and its also very quick. I doubt I'd really notice a difference with my Vertex 4 unless I tried to look for it by comparing them in identical PCs

[Arkaign]

True, the 320 was a great great drive, the Vertex 4's major boost is great write performance. I quite liked it.

I don't have the room on a 250/256 drive to run my major applications completely, the datasets are simply too large, so write speeds aren't really important to me at this time. If I get the two additional 840 Pro drives, I'll be able to move my Solidworks and Premiere tasks to them, probably keep the 840 non-pro as my OS/app drive. If I do that, I can just tell Acronis to use my old WD Black Raid array to keep the data backed up nightly from the SSDs.

[Ao1]

Originally Posted by Mfusick

I think it placebo or perhaps a fresh install vs a bloated one....


Every new install feels faster to me even when I use the same drive...

I agree a fresh install always feels faster. Read performance can also take a hit as free capacity starts to diminish and this seems to impact some SSD’s more than others.
Anyway I ran a couple of levels of Crysis 2 and monitored the disk busy time. The RAID 1 array consists of two X25-M’s. (My Samsung and Plextor drives are tied up in an OS array, so it’s not easy to use them and get comparable I/O).
There is a 12.62% difference in busy time between the RAID 1 array and the Hyper X for the “Semper FI or die” level and a 22.76% difference in busy time for the “In at the deep end” level. To temper that however the busy time difference is not something you would be likely to notice.


Uploaded with ImageShack.us